[PATCH] Hot-plug CPU Boot Changes

This patch alters the boot sequence to "plug in" each CPU, one at a
time.  You need the patch for each architecture, as well.  The
interface used to be "smp_boot_cpus()", "smp_commence()", and each
arch implemented the "maxcpus" boot arg itself.  With this patch,
it is:
  smp_prepare_cpus(maxcpus): probe for cpus and set up cpu_possible(cpu).
  __cpu_up(cpu): called *after* initcalls, for each cpu where
     cpu_possible(cpu) is true.
  smp_cpus_done(maxcpus): called after every cpu has been brought up

4 files changed, 108 insertions, 42 deletions

diff --git a/kernel/Makefile b/kernel/Makefile
index d0cad7d5115c..4834fc454271 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -17,6 +17,7 @@ obj-y     = sched.o dma.o fork.o exec_domain.o panic.o printk.o \
 	    sysctl.o capability.o ptrace.o timer.o user.o \
 	    signal.o sys.o kmod.o context.o futex.o platform.o
 
+obj-$(CONFIG_SMP) += cpu.o
 obj-$(CONFIG_UID16) += uid16.o
 obj-$(CONFIG_MODULES) += ksyms.o
 obj-$(CONFIG_PM) += pm.o
diff --git a/kernel/cpu.c b/kernel/cpu.c
new file mode 100644
index 000000000000..b1820a5536d6
--- /dev/null
+++ b/kernel/cpu.c
@@ -0,0 +1,54 @@
+/* CPU control.
+ * (C) 2001 Rusty Russell
+ * This code is licenced under the GPL.
+ */
+#include <linux/proc_fs.h>
+#include <linux/smp.h>
+#include <linux/init.h>
+#include <linux/notifier.h>
+#include <linux/sched.h>
+#include <linux/unistd.h>
+#include <asm/semaphore.h>
+
+/* This protects CPUs going up and down... */
+DECLARE_MUTEX(cpucontrol);
+
+static struct notifier_block *cpu_chain = NULL;
+
+/* Need to know about CPUs going up/down? */
+int register_cpu_notifier(struct notifier_block *nb)
+{
+	return notifier_chain_register(&cpu_chain, nb);
+}
+
+void unregister_cpu_notifier(struct notifier_block *nb)
+{
+	notifier_chain_unregister(&cpu_chain,nb);
+}
+
+int __devinit cpu_up(unsigned int cpu)
+{
+	int ret;
+
+	if ((ret = down_interruptible(&cpucontrol)) != 0) 
+		return ret;
+
+	if (cpu_online(cpu)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* Arch-specific enabling code. */
+	ret = __cpu_up(cpu);
+	if (ret != 0) goto out;
+	if (!cpu_online(cpu))
+		BUG();
+
+	/* Now call notifier in preparation. */
+	printk("CPU %u IS NOW UP!\n", cpu);
+	notifier_call_chain(&cpu_chain, CPU_ONLINE, (void *)cpu);
+
+ out:
+	up(&cpucontrol);
+	return ret;
+}
diff --git a/kernel/sched.c b/kernel/sched.c
index a54ee5a0356c..67ac32a24c1f 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -27,6 +27,8 @@
 #include <linux/completion.h>
 #include <linux/kernel_stat.h>
 #include <linux/security.h>
+#include <linux/notifier.h>
+#include <linux/delay.h>
 
 /*
  * Convert user-nice values [ -20 ... 0 ... 19 ]
@@ -1777,9 +1779,11 @@ void set_cpus_allowed(task_t *p, unsigned long new_mask)
 	migration_req_t req;
 	runqueue_t *rq;
 
+#if 0 /* FIXME: Grab cpu_lock, return error on this case. --RR */
 	new_mask &= cpu_online_map;
 	if (!new_mask)
 		BUG();
+#endif
 
 	preempt_disable();
 	rq = task_rq_lock(p, &flags);
@@ -1812,8 +1816,6 @@ out:
 	preempt_enable();
 }
 
-static __initdata int master_migration_thread;
-
 static int migration_thread(void * bind_cpu)
 {
 	int cpu = (int) (long) bind_cpu;
@@ -1825,15 +1827,7 @@ static int migration_thread(void * bind_cpu)
 	sigfillset(&current->blocked);
 	set_fs(KERNEL_DS);
 
-	/*
-	 * The first migration thread is started on the boot CPU, it
-	 * migrates the other migration threads to their destination CPUs.
-	 */
-	if (cpu != master_migration_thread) {
-		while (!cpu_rq(master_migration_thread)->migration_thread)
-			yield();
-		set_cpus_allowed(current, 1UL << cpu);
-	}
+	set_cpus_allowed(current, 1UL << cpu);
 	printk("migration_task %d on cpu=%d\n", cpu, smp_processor_id());
 	ret = setscheduler(0, SCHED_FIFO, &param);
 
@@ -1890,29 +1884,33 @@ repeat:
 	}
 }
 
-void __init migration_init(void)
+static int migration_call(struct notifier_block *nfb,
+			  unsigned long action,
+			  void *hcpu)
 {
-	int cpu;
-
-	master_migration_thread = smp_processor_id();
-	current->cpus_allowed = 1UL << master_migration_thread;
-	
-	for (cpu = 0; cpu < NR_CPUS; cpu++) {
-		if (!cpu_online(cpu))
-			continue;
-		if (kernel_thread(migration_thread, (void *) (long) cpu,
-				CLONE_FS | CLONE_FILES | CLONE_SIGNAL) < 0)
-			BUG();
+	switch (action) {
+	case CPU_ONLINE:
+		printk("Starting migration thread for cpu %li\n",
+		       (long)hcpu);
+		kernel_thread(migration_thread, hcpu,
+			      CLONE_FS | CLONE_FILES | CLONE_SIGNAL);
+		break;
 	}
-	current->cpus_allowed = -1L;
+	return NOTIFY_OK;
+}
 
-	for (cpu = 0; cpu < NR_CPUS; cpu++) {
-		if (!cpu_online(cpu))
-			continue;
-		while (!cpu_rq(cpu)->migration_thread)
-			schedule_timeout(2);
-	}
+static struct notifier_block migration_notifier = { &migration_call, NULL, 0 };
+
+int __init migration_init(void)
+{
+	/* Start one for boot CPU. */
+	migration_call(&migration_notifier, CPU_ONLINE,
+		       (void *)smp_processor_id());
+	register_cpu_notifier(&migration_notifier);
+	return 0;
 }
+
+__initcall(migration_init);
 #endif
 
 extern void init_timervecs(void);
diff --git a/kernel/softirq.c b/kernel/softirq.c
index 135106c390c4..38ae82b344c0 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -17,6 +17,7 @@
 #include <linux/init.h>
 #include <linux/tqueue.h>
 #include <linux/percpu.h>
+#include <linux/notifier.h>
 
 /*
    - No shared variables, all the data are CPU local.
@@ -387,20 +388,32 @@ static int ksoftirqd(void * __bind_cpu)
 	}
 }
 
-static __init int spawn_ksoftirqd(void)
+static int __devinit cpu_callback(struct notifier_block *nfb,
+				  unsigned long action,
+				  void *hcpu)
 {
-	int cpu;
+	int hotcpu = (unsigned long)hcpu;
 
-	for (cpu = 0; cpu < NR_CPUS; cpu++) {
-		if (!cpu_online(cpu))
-			continue;
-		if (kernel_thread(ksoftirqd, (void *) (long) cpu,
-				  CLONE_FS | CLONE_FILES | CLONE_SIGNAL) < 0)
-			printk("spawn_ksoftirqd() failed for cpu %d\n", cpu);
-		else
-			while (!ksoftirqd_task(cpu))
-				yield();
-	}
+	if (action == CPU_ONLINE) {
+		if (kernel_thread(ksoftirqd, hcpu,
+				  CLONE_FS | CLONE_FILES | CLONE_SIGNAL) < 0) {
+			printk("ksoftirqd for %i failed\n", hotcpu);
+			return NOTIFY_BAD;
+		}
+
+		while (!ksoftirqd_task(hotcpu))
+			yield();
+		return NOTIFY_OK;
+ 	}
+	return NOTIFY_BAD;
+}
+
+static struct notifier_block cpu_nfb = { &cpu_callback, NULL, 0 };
+
+static __init int spawn_ksoftirqd(void)
+{
+	cpu_callback(&cpu_nfb, CPU_ONLINE, (void *)smp_processor_id());
+	register_cpu_notifier(&cpu_nfb);
 	return 0;
 }